Google has launched public cloud infrastructure services, offering customers to spin up virtual machines hosted in its data centers.

Called Google Compute Engine, the Infrastructure-as-a-Service offering puts Google in direct competition with a lot of players in this space, including its leaders like Amazon Web Services and Rackspace.

In a post on Google’s Developers Blog, Craig McLuckie, product manager for Google Compute Engine, wrote that the company would offer Linux virtual machines on the same infrastructure that runs Google.

“This goes beyond just giving you greater flexibility and control,” McLuckie wrote. “Access to computing resources at this scale can fundamentally change the way you think about tackling a problem.”

The offering adds to Google’s previously existing Platform-as-a-Service offering Google App Engine.

Judging by a number of past high-profile outages in data centers by IaaS providers (such as the Amazon outage earlier this month), an important part of a cloud infrastructure service is providing users with the ability to build redundant geographically dispersed cloud infrastructure. This way, if a data center in Virginia, for example, has a power outage, the customer’s website hosted there can stay up because it is also hosted out of a California data center.

Google has kept tight control on information about its data center infrastructure. There are 11 data center sites the company has talked about publicly, but there may be many more.

More than half of its data centers are in the US (South Carolina, Iowa, Georgia, Oklahoma, North Carolina and Oregon). Google also has data centers in Finland, Belgium, Hong Kong, Singapore and Taiwan.

It was not clear from McLuckie’s blog or from the Compute Engine website whether users would be able to spin up VMs hosted in any of these data center locations.

Google is offering one- , two- , four- and eight-core Linux VMs, with 3.75GB of memory per core. These can be used in concert with cloud-based storage resources Google also offers as a service.

Using Google’s high-performance network technology, users can build massive compute clusters out of their cloud VMs. “You can launch enormous compute clusters - tens of thousands of cores or more,” McLuckie wrote.

Google has partnered with a number of companies that provide services on top of IaaS offerings like its Compute Engine. These partners include MapR (Apache Hadoop distribution), Cliqr (helps move applications between different kinds of clouds), Numerate (has a platform for drug development built on cloud infrastructure), Opscode (cloud automation), Puppet Labs (cloud management) and RightScale (also cloud management).